arXiv subjects

Lei Ma

Publications and source records attributed to Lei Ma.

At least 73 records · Page 4Linked to original sources

arXiv2026

S2ST-Omni: Hierarchical Language-Aware SpeechLLM Adaptation for Multilingual Speech-to-Speech Translation

Despite recent advances in speech-to-speech translation (S2ST), it remains difficult to achieve both high translation accuracy and practical flexibility. In this paper, we present S2ST-Omni, a compositional S2ST framework that integrates a high-accuracy speech-to-text translation (S2TT) frontend with a modular, plug-and-play text-to-speech (TTS) backend, enabling independent optimization of translation and synthesis. On the S2TT side, we introduce a hybrid adapter that follows a "local-then-global" strategy to bridge a pretrained Whisper encoder and a Qwen3 LLM, yielding a hierarchical acoustic-to-semantic abstraction. Building on this bridge, we further propose a hierarchical language-aware architecture that injects source-language information at two complementary levels. At the acoustic level, Language-Aware Dual-CTC operates on intermediate adapter features and employs FiLM-style feature modulation with a learnable gate, encouraging the model to learn language-specific but content-faithful acoustic representations. At the linguistic level, Language-Aware Prompting dynamically constructs source-language-conditioned prompts that activate language-specific translation knowledge in the LLM. To enable efficient optimization, we design a task-specific progressive fine-tuning strategy that first stabilizes speech-text alignment and then improves translation via LoRA on top of this converged foundation. The TTS backend remains fully modular and can be instantiated with any state-of-the-art synthesizer without retraining the S2TT frontend. Experiments on CVSS-C show that S2ST-Omni consistently achieves the best BLEU and ASR-BLEU across French, German, and Spanish to English directions, outperforming strong recent S2ST baselines.

eess.AS↗

arXiv2025

Motus: A Unified Latent Action World Model

While a general embodied agent must function as a unified system, current methods are built on isolated models for understanding, world modeling, and control. This fragmentation prevents unifying multimodal generative capabilities and hinders learning from large-scale, heterogeneous data. In this paper, we propose Motus, a unified latent action world model that leverages existing general pretrained models and rich, sharable motion information. Motus introduces a Mixture-of-Transformer (MoT) architecture to integrate three experts (i.e., understanding, video generation, and action) and adopts a UniDiffuser-style scheduler to enable flexible switching between different modeling modes (i.e., world models, vision-language-action models, inverse dynamics models, video generation models, and video-action joint prediction models). Motus further leverages the optical flow to learn latent actions and adopts a recipe with three-phase training pipeline and six-layer data pyramid, thereby extracting pixel-level "delta action" and enabling large-scale action pretraining. Experiments show that Motus achieves superior performance against state-of-the-art methods in both simulation (a +15% improvement over X-VLA and a +45% improvement over Pi0.5) and real-world scenarios(improved by +11~48%), demonstrating unified modeling of all functionalities and priors significantly benefits downstream robotic tasks.

cs.CV↗

arXiv2025

On-the-fly Large-scale 3D Reconstruction from Multi-Camera Rigs

Recent advances in 3D Gaussian Splatting (3DGS) have enabled efficient free-viewpoint rendering and photorealistic scene reconstruction. While on-the-fly extensions of 3DGS have shown promise for real-time reconstruction from monocular RGB streams, they often fail to achieve complete 3D coverage due to the limited field of view (FOV). Employing a multi-camera rig fundamentally addresses this limitation. In this paper, we present the first on-the-fly 3D reconstruction framework for multi-camera rigs. Our method incrementally fuses dense RGB streams from multiple overlapping cameras into a unified Gaussian representation, achieving drift-free trajectory estimation and efficient online reconstruction. We propose a hierarchical camera initialization scheme that enables coarse inter-camera alignment without calibration, followed by a lightweight multi-camera bundle adjustment that stabilizes trajectories while maintaining real-time performance. Furthermore, we introduce a redundancy-free Gaussian sampling strategy and a frequency-aware optimization scheduler to reduce the number of Gaussian primitives and the required optimization iterations, thereby maintaining both efficiency and reconstruction fidelity. Our method reconstructs hundreds of meters of 3D scenes within just 2 minutes using only raw multi-camera video streams, demonstrating unprecedented speed, robustness, and Fidelity for on-the-fly 3D scene reconstruction.

cs.CV↗

arXiv2025

Zoom in, Click out: Unlocking and Evaluating the Potential of Zooming for GUI Grounding

Grounding is a fundamental capability for building graphical user interface (GUI) agents. Although existing approaches rely on large-scale bounding box supervision, they still face various challenges, such as cross-platform generalization, complex layout analysis, and fine-grained element localization. In this paper, we investigate zoom as a strong yet underexplored prior for GUI grounding, and propose a training-free method, ZoomClick. By characterizing four key properties of zoom (i.e., pre-zoom, depth, shrink size, minimal crop size), we unlock its full capabilities for dynamic spatial focusing and adaptive context switching. Experiments demonstrate that our method significantly boosts the performance of both general vision-language and specialized GUI grounding models, achieving state-of-the-art results on several mainstream benchmarks; for example, UI-Venus-72B attains a 73.1% success rate on ScreenSpot-Pro. Furthermore, we present GUIZoom-Bench, a benchmark for evaluating model adaptability to zoom, aiming to inspire future research on improving zoom for further training and test-time scaling in GUI grounding tasks.

cs.CV↗

arXiv2025

Evaluating LLMs on Sequential API Call Through Automated Test Generation

By integrating tools from external APIs, Large Language Models (LLMs) have expanded their promising capabilities in a diverse spectrum of complex real-world tasks. However, testing, evaluation, and analysis of LLM tool use remain in their early stages. Most existing benchmarks rely on manually collected test cases, many of which cannot be automatically checked for semantic correctness and instead depend on static methods such as string matching. Additionally, these benchmarks often overlook the complex interactions that occur between sequential API calls, which are common in real-world applications. To fill the gap, in this paper, we introduce StateGen, an automated framework designed to generate diverse coding tasks involving sequential API interactions. StateGen combines state-machine-based API constraint solving and validation, energy-based sampling, and control-flow injection to generate executable programs. These programs are then translated into human-like natural language task descriptions through a collaboration of two LLM agents. Utilizing StateGen, we construct StateEval, a benchmark encompassing 120 verified test cases spanning across three representative scenarios: Session Service, Tensor Operation, and ElevenLabs MCP. Experimental results confirm that StateGen can effectively generate challenging and realistic API-oriented tasks, highlighting areas for improvement in current LLMs incorporating APIs.We make our framework and benchmark publicly available to support future research.

cs.SE↗

arXiv2025

Splats in Splats: Robust and Effective 3D Steganography towards Gaussian Splatting

3D Gaussian splatting (3DGS) has demonstrated impressive 3D reconstruction performance with explicit scene representations. Given the widespread application of 3DGS in 3D reconstruction and generation tasks, there is an urgent need to protect the copyright of 3DGS assets. However, existing copyright protection techniques for 3DGS overlook the usability of 3D assets, posing challenges for practical deployment. Here we describe splats in splats, the first 3DGS steganography framework that embeds 3D content in 3DGS itself without modifying any attributes. To achieve this, we take a deep insight into spherical harmonics (SH) and devise an importance-graded SH coefficient encryption strategy to embed the hidden SH coefficients. Furthermore, we employ a convolutional autoencoder to establish a mapping between the original Gaussian primitives' opacity and the hidden Gaussian primitives' opacity. Extensive experiments indicate that our method significantly outperforms existing 3D steganography techniques, with 5.31% higher scene fidelity and 3x faster rendering speed, while ensuring security, robustness, and user experience.

cs.CV↗

arXiv2025

Shape-preserving Tooth Segmentation from CBCT Images Using Deep Learning with Semantic and Shape Awareness

Background:Accurate tooth segmentation from cone beam computed tomography (CBCT) images is crucial for digital dentistry but remains challenging in cases of interdental adhesions, which cause severe anatomical shape distortion. Methods: To address this, we propose a deep learning framework that integrates semantic and shape awareness for shape-preserving segmentation. Our method introduces a target-tooth-centroid prompted multi-label learning strategy to model semantic relationships between teeth, reducing shape ambiguity. Additionally, a tooth-shape-aware learning mechanism explicitly enforces morphological constraints to preserve boundary integrity. These components are unified via multi-task learning, jointly optimizing segmentation and shape preservation. Results: Extensive evaluations on internal and external datasets demonstrate that our approach significantly outperforms existing methods. Conclusions: Our approach effectively mitigates shape distortions and providing anatomically faithful tooth boundaries.

cs.CV↗

arXiv2025

Seeing the Unseen in Low-light Spike Streams

Spike camera, a type of neuromorphic sensor with high-temporal resolution, shows great promise for high-speed visual tasks. Unlike traditional cameras, spike camera continuously accumulates photons and fires asynchronous spike streams. Due to unique data modality, spike streams require reconstruction methods to become perceptible to the human eye. However, lots of methods struggle to handle spike streams in low-light high-speed scenarios due to severe noise and sparse information. In this work, we propose Diff-SPK, a diffusion-based reconstruction method. Diff-SPK effectively leverages generative priors to supplement texture information under diverse low-light conditions. Specifically, it first employs an Enhanced Texture from Inter-spike Interval (ETFI) to aggregate sparse information from low-light spike streams. Then, the encoded ETFI by a suitable encoder serve as the input of ControlNet for high-speed scenes generation. To improve the quality of results, we introduce an ETFI-based feature fusion module during the generation process.

cs.CV↗

arXiv2025

DepthVanish: Optimizing Adversarial Interval Structures for Stereo-Depth-Invisible Patches

Stereo depth estimation is a critical task in autonomous driving and robotics, where inaccuracies (such as misidentifying nearby objects as distant) can lead to dangerous situations. Adversarial attacks against stereo depth estimation can help reveal vulnerabilities before deployment. Previous works have shown that repeating optimized textures can effectively mislead stereo depth estimation in digital settings. However, our research reveals that these naively repeated textures perform poorly in physical implementations, i.e., when deployed as patches, limiting their practical utility for stress-testing stereo depth estimation systems. In this work, for the first time, we discover that introducing regular intervals among the repeated textures, creating a grid structure, significantly enhances the patch's attack performance. Through extensive experimentation, we analyze how variations of this novel structure influence the adversarial effectiveness. Based on these insights, we develop a novel stereo depth attack that jointly optimizes both the interval structure and texture elements. Our generated adversarial patches can be inserted into any scenes and successfully attack advanced stereo depth estimation methods of different paradigms, i.e., RAFT-Stereo and STTR. Most critically, our patch can also attack commercial RGB-D cameras (Intel RealSense) in real-world conditions, demonstrating their practical relevance for security assessment of stereo systems. The code is officially released at: https://github.com/WiWiN42/DepthVanish

cs.CV↗

arXiv2025

DETree: DEtecting Human-AI Collaborative Texts via Tree-Structured Hierarchical Representation Learning

Detecting AI-involved text is essential for combating misinformation, plagiarism, and academic misconduct. However, AI text generation includes diverse collaborative processes (AI-written text edited by humans, human-written text edited by AI, and AI-generated text refined by other AI), where various or even new LLMs could be involved. Texts generated through these varied processes exhibit complex characteristics, presenting significant challenges for detection. Current methods model these processes rather crudely, primarily employing binary classification (purely human vs. AI-involved) or multi-classification (treating human-AI collaboration as a new class). We observe that representations of texts generated through different processes exhibit inherent clustering relationships. Therefore, we propose DETree, a novel approach that models the relationships among different processes as a Hierarchical Affinity Tree structure, and introduces a specialized loss function that aligns text representations with this tree. To facilitate this learning, we developed RealBench, a comprehensive benchmark dataset that automatically incorporates a wide spectrum of hybrid texts produced through various human-AI collaboration processes. Our method improves performance in hybrid text detection tasks and significantly enhances robustness and generalization in out-of-distribution scenarios, particularly in few-shot learning conditions, further demonstrating the promise of training-based approaches in OOD settings. Our code and dataset are available at https://github.com/heyongxin233/DETree.

cs.CL↗

arXiv2025

NANO3D: A Training-Free Approach for Efficient 3D Editing Without Masks

3D object editing is essential for interactive content creation in gaming, animation, and robotics, yet current approaches remain inefficient, inconsistent, and often fail to preserve unedited regions. Most methods rely on editing multi-view renderings followed by reconstruction, which introduces artifacts and limits practicality. To address these challenges, we propose Nano3D, a training-free framework for precise and coherent 3D object editing without masks. Nano3D integrates FlowEdit into TRELLIS to perform localized edits guided by front-view renderings, and further introduces region-aware merging strategies, Voxel/Slat-Merge, which adaptively preserve structural fidelity by ensuring consistency between edited and unedited areas. Experiments demonstrate that Nano3D achieves superior 3D consistency and visual quality compared with existing methods. Based on this framework, we construct the first large-scale 3D editing datasets Nano3D-Edit-100k, which contains over 100,000 high-quality 3D editing pairs. This work addresses long-standing challenges in both algorithm design and data availability, significantly improving the generality and reliability of 3D editing, and laying the groundwork for the development of feed-forward 3D editing models. Project Page:https://jamesyjl.github.io/Nano3D

cs.CV↗

arXiv2025

TRUSTVIS: A Multi-Dimensional Trustworthiness Evaluation Framework for Large Language Models

As Large Language Models (LLMs) continue to revolutionize Natural Language Processing (NLP) applications, critical concerns about their trustworthiness persist, particularly in safety and robustness. To address these challenges, we introduce TRUSTVIS, an automated evaluation framework that provides a comprehensive assessment of LLM trustworthiness. A key feature of our framework is its interactive user interface, designed to offer intuitive visualizations of trustworthiness metrics. By integrating well-known perturbation methods like AutoDAN and employing majority voting across various evaluation methods, TRUSTVIS not only provides reliable results but also makes complex evaluation processes accessible to users. Preliminary case studies on models like Vicuna-7b, Llama2-7b, and GPT-3.5 demonstrate the effectiveness of our framework in identifying safety and robustness vulnerabilities, while the interactive interface allows users to explore results in detail, empowering targeted model improvements. Video Link: https://youtu.be/k1TrBqNVg8g

cs.SE↗

arXiv2025

Exploring Representation Invariance in Finetuning

Foundation models pretrained on large-scale natural images are widely adapted to various cross-domain low-resource downstream tasks, benefiting from generalizable and transferable patterns captured by their representations. However, these representations are later found to gradually vanish during finetuning, accompanied by a degradation of model's original generalizability. In this paper, we argue that such tasks can be effectively adapted without sacrificing the benefits of pretrained representations. We approach this by introducing \textit{Representation Invariance FineTuning (RIFT)}, a regularization that maximizes the representation similarity between pretrained and finetuned models by leveraging orthogonal invariance of manifolds in a computationally efficient way. Experiments demonstrate that our method is compatible with mainstream finetuning methods, offering competitive or even enhanced performance and better preservation of the generalizability.

cs.CV↗

arXiv2025

Development and validation of an AI foundation model for endoscopic diagnosis of esophagogastric junction adenocarcinoma: a cohort and deep learning study

The early detection of esophagogastric junction adenocarcinoma (EGJA) is crucial for improving patient prognosis, yet its current diagnosis is highly operator-dependent. This paper aims to make the first attempt to develop an artificial intelligence (AI) foundation model-based method for both screening and staging diagnosis of EGJA using endoscopic images. In this cohort and learning study, we conducted a multicentre study across seven Chinese hospitals between December 28, 2016 and December 30, 2024. It comprises 12,302 images from 1,546 patients; 8,249 of them were employed for model training, while the remaining were divided into the held-out (112 patients, 914 images), external (230 patients, 1,539 images), and prospective (198 patients, 1,600 images) test sets for evaluation. The proposed model employs DINOv2 (a vision foundation model) and ResNet50 (a convolutional neural network) to extract features of global appearance and local details of endoscopic images for EGJA staging diagnosis. Our model demonstrates satisfactory performance for EGJA staging diagnosis across three test sets, achieving an accuracy of 0.9256, 0.8895, and 0.8956, respectively. In contrast, among representative AI models, the best one (ResNet50) achieves an accuracy of 0.9125, 0.8382, and 0.8519 on the three test sets, respectively; the expert endoscopists achieve an accuracy of 0.8147 on the held-out test set. Moreover, with the assistance of our model, the overall accuracy for the trainee, competent, and expert endoscopists improves from 0.7035, 0.7350, and 0.8147 to 0.8497, 0.8521, and 0.8696, respectively. To our knowledge, our model is the first application of foundation models for EGJA staging diagnosis and demonstrates great potential in both diagnostic accuracy and efficiency.

cs.CV↗

arXiv2025

Digging Into the Internal: Causality-Based Analysis of LLM Function Calling

Function calling (FC) has emerged as a powerful technique for facilitating large language models (LLMs) to interact with external systems and perform structured tasks. However, the mechanisms through which it influences model behavior remain largely under-explored. Besides, we discover that in addition to the regular usage of FC, this technique can substantially enhance the compliance of LLMs with user instructions. These observations motivate us to leverage causality, a canonical analysis method, to investigate how FC works within LLMs. In particular, we conduct layer-level and token-level causal interventions to dissect FC's impact on the model's internal computational logic when responding to user queries. Our analysis confirms the substantial influence of FC and reveals several in-depth insights into its mechanisms. To further validate our findings, we conduct extensive experiments comparing the effectiveness of FC-based instructions against conventional prompting methods. We focus on enhancing LLM safety robustness, a critical LLM application scenario, and evaluate four mainstream LLMs across two benchmark datasets. The results are striking: FC shows an average performance improvement of around 135% over conventional prompting methods in detecting malicious inputs, demonstrating its promising potential to enhance LLM reliability and capability in practical applications.

cs.SE↗

arXiv2025

Strong Correlation Driven Quadrupolar to Dipolar Exciton Transitions in a Trilayer Moiré Superlattice

The additional layer degree of freedom in trilayer moiré superlattices of transition metal dichalcogenides enables the emergence of novel excitonic species, such as quadrupolar excitons, which exhibit unique excitonic interactions and hold promise for realizing intriguing excitonic phases and their quantum phase transitions. Concurrently, the presence of strong electronic correlations in moiré superlattices, as exemplified by the observations of Mott insulators and generalized Wigner crystals, offers a direct route to manipulate these new excitonic states and resulting collective excitonic phases. Here, we demonstrate that strong exciton-exciton and electron-exciton interactions, both stemming from robust electron correlations, can be harnessed to controllably drive transitions between quadrupolar and dipolar excitons. This is achieved by tuning either the exciton density or electrostatic doping in a trilayer semiconducting moiré superlattice. Our findings not only advance the fundamental understanding of quadrupolar excitons but also usher in new avenues for exploring and engineering many-body quantum phenomena through novel correlated excitons in semiconducting moiré systems.

cond-mat.mes-hall↗

arXiv2025

Aligning Instruction Tuning with Pre-training

Instruction tuning enhances large language models (LLMs) to follow human instructions across diverse tasks, relying on high-quality datasets to guide behavior. However, these datasets, whether manually curated or synthetically generated, are often narrowly focused and misaligned with the broad distributions captured during pre-training, limiting LLM generalization and effective use of pre-trained knowledge. We propose Aligning Instruction Tuning with Pre-training (AITP), a method that bridges this gap by identifying coverage shortfalls in instruction-tuning datasets and rewriting underrepresented pre-training data into high-quality instruction-response pairs. This approach enriches dataset diversity while preserving task-specific objectives. Evaluations on three fully open LLMs across eight benchmarks demonstrate consistent performance improvements with AITP. Ablations highlight the benefits of adaptive data selection, controlled rewriting, and balanced integration, emphasizing the importance of aligning instruction tuning with pre-training distributions to unlock the full potential of LLMs.

cs.AI↗

arXiv2025

Zero-Shot Voice Conversion via Content-Aware Timbre Ensemble and Conditional Flow Matching

Despite recent advances in zero-shot voice conversion (VC), achieving speaker similarity and naturalness comparable to ground-truth recordings remains a significant challenge. In this letter, we propose CTEFM-VC, a zero-shot VC framework that integrates content-aware timbre ensemble modeling with conditional flow matching. Specifically, CTEFM-VC decouples utterances into content and timbre representations and leverages a conditional flow matching model to reconstruct the Mel-spectrogram of the source speech. To enhance its timbre modeling capability and naturalness of generated speech, we first introduce a context-aware timbre ensemble modeling approach that adaptively integrates diverse speaker verification embeddings and enables the effective utilization of source content and target timbre elements through a cross-attention module. Furthermore, a structural similarity-based timbre loss is presented to jointly train CTEFM-VC end-to-end. Experiments show that CTEFM-VC consistently achieves the best performance in all metrics assessing speaker similarity, speech naturalness, and intelligibility, significantly outperforming state-of-the-art zero-shot VC systems.

cs.SD↗